Published June 4, 2026
| Version v1
Dataset
Restricted
WAMI — Computed Evo2-7B block embeddings, HAPLOFM kinship matrix, and consensus sequences for genome-wide wheat discovery
Authors/Creators
- 1. Georg-August-Universitaet Goettingen
- 2. Syngenta, Jealotts Hill International Research Centre
- 3. CIMMYT
Description
Computed data for: A frozen DNA foundation model read as relatedness extends a validated wheat marker panel into genome-wide discovery.
Contents:
1. Per-cultivar per-block Evo2-7B embeddings (285x1752x4096, z-scored)
2. 285x285 HAPLOFM block-cosine kinship matrix
3. Compact embeddings NPZ (274 MB)
4. 8192-bp consensus sequences per (cultivar, block)
Code: https://github.com/KhasimHussainBajiShaik/WAMI
Base model: arcinstitute/evo2_7b_base (Brixi et al., Nature 2026, DOI: 10.1038/s41586-026-10176-5)
ACCESS NOTE: Files are restricted pending paper publication. Metadata is visible; files will be made openly available upon acceptance.
Files
Additional details
Related works
- Is supplement to
- Software: https://github.com/KhasimHussainBajiShaik/WAMI (URL)
- References
- Journal article: 10.1038/s41586-026-10176-5 (DOI)